US20040001607A1 - Method of designing watermark in consideration of wiener attack and whitening filtered detection - Google Patents
Method of designing watermark in consideration of wiener attack and whitening filtered detection Download PDFInfo
- Publication number
- US20040001607A1 US20040001607A1 US10/271,530 US27153002A US2004001607A1 US 20040001607 A1 US20040001607 A1 US 20040001607A1 US 27153002 A US27153002 A US 27153002A US 2004001607 A1 US2004001607 A1 US 2004001607A1
- Authority
- US
- United States
- Prior art keywords
- watermark
- spectral density
- power spectral
- wiener
- attack
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
- H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
- H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
- H04N1/32149—Methods relating to embedding, encoding, decoding, detection or retrieval operations
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
- H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
- H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
- H04N1/32149—Methods relating to embedding, encoding, decoding, detection or retrieval operations
- H04N1/3232—Robust embedding or watermarking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0052—Embedding of the watermark in the frequency domain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0065—Extraction of an embedded watermark; Reliable detection
Definitions
- the present invention relates to a watermark, and more particularly, to a method of designing a watermark having the power spectral density optimized so that the performance of detecting watermark can be improved by employing the whitening filtered detection after the Wiener attack.
- the watermark should not be perceivable by the human visual system, the human auditory system and so on.
- the energy of the watermark should be small enough so that the watermark cannot be perceived.
- the watermark should be robust so as to resist intentional or unintentional attack.
- the unintentional attack means a loss of information in common signal processing such as lossy compression, low-pass filtering, scaling, digital-to-analog or analog-to-digital conversion, resampling, etc.
- the intentional attack means a removal operation of the watermark within preserving the quality of the digital contents.
- an algorithm for embedding the watermark should be secure enough so that the embedded watermark cannot be removed even if the algorithm becomes known. That is, only the copyright holder can access the watermark.
- a well-known power spectrum condition is derived as a defense against the Wiener attack.
- the Wiener attack means estimation and extraction of the embedded watermark from the signal including the watermark for the purpose of obstructing the detection of the watermark.
- the power spectral density of the watermark should be in proportion to the power spectral density of the original signal. This is because it is difficult to estimate the watermark from the signal including the watermark in case of conforming to the power spectrum condition.
- the power spectrum condition considers only the detection based on a correlation, but does not consider employing the whitening filter prior to the detection.
- FIG. 1 is a graph illustrating the power spectral density of the watermark and the original signal according to a conventional method.
- the power spectral density of the watermark is designed to be in proportion to the power spectral density of the original signal. This proportion of the power spectral densities makes the estimation of the watermark due to an invasion and so on difficult.
- the whitening filter is not considered.
- the robustness and the imperceptibility of the watermark can be represented by an L2 efficiency, which indicates the energy of the watermark that the attacker can remove.
- the article shows that the power spectrum condition should be satisfied for the maximum L2 efficiency. That is, the power spectrum of the watermark should be the one obtained by reducing the power spectrum of the original signal in a specified ratio.
- a watermark that satisfies the power spectrum condition is the most efficient one among the watermarks independent from the original signal. In the analysis of one-dimensional and two-dimensional signal models and experiments using actual images, the watermark that satisfies the power spectrum condition shows an excellent performance.
- U.S. Pat. No. 5,848,155 entitled “Spread Spectrum Watermark for Embedded Signaling”
- a watermark is embedded into audio, video and image multimedia data using spread spectrum methodology.
- This patent overcomes the limitations of the prior systems using a spread spectrum watermarking method that embeds a watermark or certain information into perceptually significant regions of an image.
- the watermark can be extracted from the watermarked data without using the original data.
- the preferred method of watermarking extraction is to use a spatial or temporal local average of the frequency coefficients of the watermarked data to determine the watermark.
- the frequency coefficients of a two-dimensional neighborhood in two-dimensional watermarked data i.e., an image
- the Wiener attack means to estimate the watermark using the Wiener filter and to obstruct the detection by subtracting the estimated watermark from the watermarked data.
- the robustness and the imperceptibility of the watermark can be measured by the energy of the estimated watermark that can be removed by the Wiener attack.
- the most efficient watermark should satisfy the power spectrum condition. That is, the power spectrum of the watermark is obtained by reducing the power spectrum of the original signal in a specified ratio.
- the power spectrum condition compliant watermark shows the optimal performance against the Wiener attack.
- the whitening filter provides no gain for the power spectrum compliant watermark.
- the present invention is directed to a method of designing a watermark in consideration of the Wiener attack and the whitening filtered detection that substantially obviates one or more problems due to the limitations and the drawbacks of the related art.
- the present invention is for designing the power spectral density of the watermark using an optimization method so that the entire performance of the detection can be improved by reflecting the gain of the whitening filter after the Wiener attack.
- the present invention is for obtaining a higher detection gain using the whitening filter after the Wiener attack, and maximizing the expected value of the difference between the test statistics of the hypotheses that the watermark exists and the watermark does not exist, respectively, to optimize the detection performance.
- the expected value of the difference between the test statistics as an objective function, the power spectral density of the watermark, which corresponds to a maximum differentiated value of the power spectral density of the watermark using the Lagrange multiplier method, is obtained.
- the energy of the watermark is given as a condition, so the scale factor of the watermark can be adjusted to match the energy.
- an object of the present invention is to provide a method of designing a watermark that makes a strong defense against the Wiener attack and considers the whitening filtered detection.
- ⁇ x ( ⁇ ) is the power spectral density of the original signal
- ⁇ w ( ⁇ ) is the power spectral density of the watermark
- ⁇ is a frequency within the range of [ ⁇ , ⁇ ).
- the optimized power spectrum of the watermark as shown in the following equation is obtained from the objective function using the Lagrange multiplier method.
- ⁇ w ⁇ ( ⁇ ) - ⁇ x ⁇ ( ⁇ ) + ⁇ x ⁇ ( ⁇ ) 1 2 ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ - ⁇ ⁇ ⁇ ⁇ x ⁇ ( ⁇ ) ⁇ ⁇ ⁇ ⁇ ( ⁇ x 2 + ⁇ w 2 ) ,
- ⁇ x 2 is the variance of the original signal
- ⁇ w 2 is the variance of the watermark
- FIG. 1 is a graph illustrating the power spectral density of a watermark and an original signal according to a conventional method.
- FIG. 2 is a block diagram of a terminal that can perform a watermark in consideration of the whitening filtered detection according to the present invention.
- FIG. 3 is a block diagram illustrating the Wiener attack and the whitening filter.
- FIG. 6 is a graph of the power spectral density ⁇ x ( ⁇ ) of the original signal.
- FIG. 7 is a graph of the power spectral density ⁇ w ( ⁇ ) of the watermark according to the present invention.
- FIG. 8 is a graph illustrating the correlation of the power spectral densities of the watermark and the original signal according to the present invention when the optimal result is obtained.
- FIG. 2 is a block diagram of a terminal that can perform watermarking in consideration of the whitening filtered detection according to the present invention.
- the terminal is provided with a central processing unit 11 , a main memory 12 connected to the central processing unit 11 , an auxiliary memory 13 connected to the main memory 12 , and input and output devices 10 and 14 connected to the central processing unit 11 .
- the central processing unit 11 controls and manages the whole operation of the terminal, and the main memory 12 and the auxiliary memory 13 store programs to be performed by the central processing unit 11 and various kinds of data used or produced during the operation of the terminal.
- the input device 10 and the output device are for data input/output with a user.
- the auxiliary memory 13 serves to store a large capacity of data.
- the input device 10 includes user input means such as a general keyboard, mouse, tablet, touch screen, etc, and the output device 14 includes a display, printer, video recorder, etc.
- the original signal i.e., an image, audio, etc.
- the noise generally has a colored power spectrum.
- the detection performance can be greatly improved. If the power spectral density of the watermark is equal to the power spectral density of the original signal when using the whitening filter, the gain of the whitening filter becomes 1, and the detection performance doesn't improve although the watermark can resist quite effectively the Wiener attack. On the contrary, when the power spectral density of the watermark is different from the power spectral density of the original signal, the watermark can be removed by the Wiener attack. In spite of that, the detection performance does improve because the gain of the whitening filter increases.
- the present invention is for designing a watermark that is robust to the Wiener attack, considers the gain of the whitening filter, and thus has an improved detection performance.
- the watermark w[n] of the present invention is a N -dimensional wide-sense stationary (WSS) Gaussian random process with zero mean, and variance ⁇ w 2 , and interacts with variables such as the power spectral density ⁇ w ( ⁇ ).
- WSS wide-sense stationary
- the original signal x[n] has variance ⁇ x 2 and a power spectral density ⁇ x ( ⁇ ). Meanwhile, ⁇ is a frequency within the range of [ ⁇ , ⁇ ).
- the embedding of the watermark is performed in a manner that the watermark signal w[n] is added to the original signal x[n].
- a receiver detects the watermark by calculating a correlation.
- the original signal x[n] is considered as a colored noise, and the detection performance is greatly improved by the whitening filter.
- R x is the covariance matrix of the original signal x[n].
- [0054] is selected H 1 .
- ⁇ is the detection threshold, and is decided by selection.
- the Wiener attack is the removal of the estimated watermark, where the estimation is performed from the watermarked signal, in order to prevent the detection of the watermark.
- h[n] is the impulse response of the Wiener filter
- ⁇ ( ⁇ 0) is a scalar gain factor
- the present invention provides a watermark designing method that satisfies the two conditions.
- FIG. 3 is a block diagram illustrating the Wiener attack and the whitening filter.
- y[n] denotes a received signal including the watermark w[n] and the original signal x[n], g[n] the impulse response of the whitening filter after the Wiener attack, and w[n] the watermark signal.
- a correlation detector performs a detection based on correlation between the two signals. After the Wiener attack, the signal passes through the whitening filter g[n], and then existence/nonexistence of the watermark is finally discriminated through the correlation detector.
- the watermark is designed so as to maximize the power spectral density ⁇ x ( ⁇ ) of the original signal and the power spectral density ⁇ w ( ⁇ ) of the watermark in the frequency range of [ ⁇ , ⁇ ).
- the power spectral density of the watermark is first extracted using the Lagrange multiplier method.
- g( ⁇ w ( ⁇ )) is an integrand.
- the power spectral density ⁇ w ( ⁇ ) of the watermark should be a positive real number.
- the power spectral density having a negative value is removed.
- ⁇ w 2 - ⁇ x 2 + 1 2 ⁇ ⁇ ⁇ ⁇ - ⁇ ⁇ ⁇ - ⁇ ⁇ ⁇ ⁇ x ⁇ ( ⁇ ) ⁇ ⁇ ⁇ .
- ⁇ - ( ⁇ - ⁇ ⁇ ⁇ ⁇ x ⁇ ( ⁇ ) ⁇ ⁇ ⁇ 2 ⁇ ⁇ ⁇ ⁇ ( ⁇ w 2 + ⁇ x 2 ) ) 2 .
- the optimized power spectral density of the watermark can be obtained using the Lagrange multiplier method.
- ⁇ w ⁇ ( ⁇ ) - ⁇ x ⁇ ( ⁇ ) + ⁇ x ⁇ ( ⁇ ) 1 2 ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ - ⁇ ⁇ ⁇ ⁇ x ⁇ ( ⁇ ) ⁇ ⁇ ⁇ ⁇ ⁇ ( ⁇ x 2 + ⁇ w 2 ) ( 8 )
- the above equation (8) shows the power spectral density of the designed watermark w[n], and indicates that the power spectral density ⁇ w ( ⁇ ) of the watermark is not in proportion to the power spectral density ⁇ x ( ⁇ ) of the original signal. Also, the equation (8) represents a convex curve. If ⁇ square root ⁇ square root over ( ⁇ x ( ⁇ )) ⁇ is too large, the power spectral density ⁇ w ( ⁇ ) of the watermark is reduced on the contrary, and thus if the large power spectral density ⁇ w ( ⁇ ) of the watermark is in proportion to the power spectral density ⁇ x ( ⁇ ) of the original signal, the operating efficiency is not good.
- WDR watermark document ratio
- FIGS. 4 and 5 the optimum watermark corresponds to the watermark according to the present invention.
- FIGS. 4 and 5 illustrate the receiver operating characteristics, and show the relation between a probability of false alarm and a normal probability of detection. Under the same probability of false alarm, the performance becomes better as the probability of detection becomes higher. A better performance is shown under the ratio of the watermark to the original signal of ⁇ 20 dB and ⁇ 30 dB. As shown in the drawings, the WDR of the watermark designed by the optimization method between ⁇ 20 dB and ⁇ 30 dB represents an improved detection performance.
- FIG. 6 is a graph of a power spectral density ⁇ x ( ⁇ ) of the original signal
- FIG. 7 is a graph of a power spectral density ⁇ w ( ⁇ ) of the watermark according to the present invention.
- the power spectral density of the watermark should be in high-frequency regions. However, as shown in FIG. 7, even if the power spectral density of the watermark is concentrated in low-frequency regions, a good detection performance is shown after the whitening filtering.
- the watermark having the optimum condition according to the present invention is obtained by the following equation.
- T ⁇ x 2 + ⁇ w 2 - ⁇ x ⁇ ( ⁇ ) 2 ⁇ ⁇ ⁇ ⁇ ⁇ - ⁇ ⁇ ⁇ ⁇ x ⁇ ( ⁇ ) ⁇ ⁇ ⁇ ⁇ ⁇ 0 ( 10 )
- T is the optimized test statistics.
- FIG. 8 illustrates the watermark according to the present invention showing the optimum result. There is not a great difference between the power spectral density of the watermark designed according to the present invention and the power spectral density of the original signal, but the performance of the whitening filtered detection improves.
- the watermark designed according to the present invention shows an improved detection performance in a detection environment using the whitening filter after the Wiener attack.
- the gain of the whitening filter becomes 1, and thus the detection performance is not improved. Also, if the power spectral density of the watermark is opposite to the power spectral density of the original signal, the watermark is easily removed by the Wiener attack even if the whitening filter has any gain, and this deteriorates the detection performance of the whitening filter. On the contrary, according to the present invention, the detection performance is improved using the watermark having the optimized power spectral density in consideration of both the Wiener attack and the whitening filter.
Abstract
Description
- 1. Field of the Invention
- The present invention relates to a watermark, and more particularly, to a method of designing a watermark having the power spectral density optimized so that the performance of detecting watermark can be improved by employing the whitening filtered detection after the Wiener attack.
- 2. Background of the Related Art
- Generally, digital media such as audio, image and video have been rapidly increased. Since such digital media products can be easily distributed, forged and copied, it is necessary to protect their copyrights. Digital watermarking has been appeared as a popular means for satisfying such necessity. A digital watermark is hidden information permanently remaining in a digital signal without being perceived by humans. This watermark has information related to an owner, distributor and receiver of the digital signal. When the owner of this digital type work is inquired, it can be recognized through the extraction of inserted information.
- As a security means for strengthening the copyright, the watermark should not be perceivable by the human visual system, the human auditory system and so on. Thus, the energy of the watermark should be small enough so that the watermark cannot be perceived. Also, the watermark should be robust so as to resist intentional or unintentional attack.
- The unintentional attack means a loss of information in common signal processing such as lossy compression, low-pass filtering, scaling, digital-to-analog or analog-to-digital conversion, resampling, etc. The intentional attack means a removal operation of the watermark within preserving the quality of the digital contents.
- Meanwhile, an algorithm for embedding the watermark should be secure enough so that the embedded watermark cannot be removed even if the algorithm becomes known. That is, only the copyright holder can access the watermark.
- The most important thing is that the copyright holder can be accurately discriminated by the watermark.
- A well-known power spectrum condition is derived as a defense against the Wiener attack. The Wiener attack means estimation and extraction of the embedded watermark from the signal including the watermark for the purpose of obstructing the detection of the watermark. According to the power spectrum condition, the power spectral density of the watermark should be in proportion to the power spectral density of the original signal. This is because it is difficult to estimate the watermark from the signal including the watermark in case of conforming to the power spectrum condition. However, the power spectrum condition considers only the detection based on a correlation, but does not consider employing the whitening filter prior to the detection.
- FIG. 1 is a graph illustrating the power spectral density of the watermark and the original signal according to a conventional method.
- According to the conventional method, the power spectral density of the watermark is designed to be in proportion to the power spectral density of the original signal. This proportion of the power spectral densities makes the estimation of the watermark due to an invasion and so on difficult. However, in the conventional watermark design, the whitening filter is not considered.
- Meanwhile, in an article by Jonathan K. Su and Bernd Girod, entitled “On the Robustness and Imperceptibility of Digital Fingerprints”, Proc. IEEE Conf. on Multimedia Computing and Systems, pp. 530-535, vol. 2, June 1999, it is described that in most digital watermarking applications, the most preferable characteristics are robustness and imperceptibility. This article shows a theoretical analysis of the watermarking. Here, an attacker estimates the watermark using the Wiener filter.
- The robustness and the imperceptibility of the watermark can be represented by an L2 efficiency, which indicates the energy of the watermark that the attacker can remove. The article shows that the power spectrum condition should be satisfied for the maximum L2 efficiency. That is, the power spectrum of the watermark should be the one obtained by reducing the power spectrum of the original signal in a specified ratio. A watermark that satisfies the power spectrum condition is the most efficient one among the watermarks independent from the original signal. In the analysis of one-dimensional and two-dimensional signal models and experiments using actual images, the watermark that satisfies the power spectrum condition shows an excellent performance.
- Also, in U.S. Pat. No. 5,848,155, entitled “Spread Spectrum Watermark for Embedded Signaling”, a watermark is embedded into audio, video and image multimedia data using spread spectrum methodology. This patent overcomes the limitations of the prior systems using a spread spectrum watermarking method that embeds a watermark or certain information into perceptually significant regions of an image. Specifically, according to this patent, the watermark can be extracted from the watermarked data without using the original data. The preferred method of watermarking extraction is to use a spatial or temporal local average of the frequency coefficients of the watermarked data to determine the watermark. For instance, the frequency coefficients of a two-dimensional neighborhood in two-dimensional watermarked data, i.e., an image, are analyzed to reproduce the entire watermark. This is possible since the watermark is embedded into the data using spread spectrum technology that spreads the watermark throughout the data.
- Meanwhile, the Wiener attack means to estimate the watermark using the Wiener filter and to obstruct the detection by subtracting the estimated watermark from the watermarked data. The robustness and the imperceptibility of the watermark can be measured by the energy of the estimated watermark that can be removed by the Wiener attack. Regarding the energy of the estimated watermark, the most efficient watermark should satisfy the power spectrum condition. That is, the power spectrum of the watermark is obtained by reducing the power spectrum of the original signal in a specified ratio. In case of using the matched filter only, the power spectrum condition compliant watermark shows the optimal performance against the Wiener attack. However, in case of employing the whitening filter prior to the detection, the whitening filter provides no gain for the power spectrum compliant watermark.
- Accordingly, the present invention is directed to a method of designing a watermark in consideration of the Wiener attack and the whitening filtered detection that substantially obviates one or more problems due to the limitations and the drawbacks of the related art. The present invention is for designing the power spectral density of the watermark using an optimization method so that the entire performance of the detection can be improved by reflecting the gain of the whitening filter after the Wiener attack.
- Also, the present invention is for obtaining a higher detection gain using the whitening filter after the Wiener attack, and maximizing the expected value of the difference between the test statistics of the hypotheses that the watermark exists and the watermark does not exist, respectively, to optimize the detection performance. Specifically, regarding the expected value of the difference between the test statistics as an objective function, the power spectral density of the watermark, which corresponds to a maximum differentiated value of the power spectral density of the watermark using the Lagrange multiplier method, is obtained. Here, the energy of the watermark is given as a condition, so the scale factor of the watermark can be adjusted to match the energy.
- As described above, an object of the present invention is to provide a method of designing a watermark that makes a strong defense against the Wiener attack and considers the whitening filtered detection.
- Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
- To achieve these objects and other advantages and in accordance with the purpose of the invention, as embodied and broadly described herein, there is provided a method of designing a watermark that improves the detection performance by maximizing the expected value of the difference between the test statistics of the two hypotheses where the watermark exists and where the watermark does not exist, respectively, in the following equation.
-
- where Φx(Ω) is the power spectral density of the original signal, Φw(Ω) is the power spectral density of the watermark, and Ω is a frequency within the range of [−π,π).
-
- where σx 2 is the variance of the original signal, and σw 2 is the variance of the watermark.
- It is to be understood that both the foregoing general description and the following detailed description of the present invention are exemplary and explanatory and are intended to provide further explanation of the invention as claimed.
- The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principle of the invention. In the drawings:
- FIG. 1 is a graph illustrating the power spectral density of a watermark and an original signal according to a conventional method.
- FIG. 2 is a block diagram of a terminal that can perform a watermark in consideration of the whitening filtered detection according to the present invention.
- FIG. 3 is a block diagram illustrating the Wiener attack and the whitening filter.
- FIG. 4 is a graph illustrating receiver operating characteristics in case that WDR=−20 dB.
- FIG. 5 is a graph illustrating receiver operating characteristics in case that WDR=−30 dB.
- FIG. 6 is a graph of the power spectral density Φx(Ω) of the original signal.
- FIG. 7 is a graph of the power spectral density Φw(Ω) of the watermark according to the present invention.
- FIG. 8 is a graph illustrating the correlation of the power spectral densities of the watermark and the original signal according to the present invention when the optimal result is obtained.
- The method of designing a watermark according to the preferred embodiment of the present invention will now be explained in detail with reference to the accompanying drawings.
- FIG. 2 is a block diagram of a terminal that can perform watermarking in consideration of the whitening filtered detection according to the present invention.
- The terminal is provided with a
central processing unit 11, amain memory 12 connected to thecentral processing unit 11, anauxiliary memory 13 connected to themain memory 12, and input andoutput devices central processing unit 11. - The
central processing unit 11 controls and manages the whole operation of the terminal, and themain memory 12 and theauxiliary memory 13 store programs to be performed by thecentral processing unit 11 and various kinds of data used or produced during the operation of the terminal. Theinput device 10 and the output device are for data input/output with a user. - The
auxiliary memory 13 serves to store a large capacity of data. Theinput device 10 includes user input means such as a general keyboard, mouse, tablet, touch screen, etc, and theoutput device 14 includes a display, printer, video recorder, etc. - In detecting the watermark, the original signal (i.e., an image, audio, etc.) is considered as a noise, and generally has a colored power spectrum.
- Accordingly, by applying the whitening filter to the watermark detection operation, the detection performance can be greatly improved. If the power spectral density of the watermark is equal to the power spectral density of the original signal when using the whitening filter, the gain of the whitening filter becomes 1, and the detection performance doesn't improve although the watermark can resist quite effectively the Wiener attack. On the contrary, when the power spectral density of the watermark is different from the power spectral density of the original signal, the watermark can be removed by the Wiener attack. In spite of that, the detection performance does improve because the gain of the whitening filter increases.
- As described above, the present invention is for designing a watermark that is robust to the Wiener attack, considers the gain of the whitening filter, and thus has an improved detection performance.
- In order for the watermark according to the present invention to be robust to the Wiener attack and to show an improved performance in the whitening filtered detection, a new modified condition according to the present invention should be satisfied instead of the conventional power spectrum condition (PSC).
- The watermark w[n] of the present invention is a N -dimensional wide-sense stationary (WSS) Gaussian random process with zero mean, and variance σw 2, and interacts with variables such as the power spectral density Φw(Ω).
- In the same manner, the original signal x[n] has variance σx 2 and a power spectral density Φx(Ω). Meanwhile, Ω is a frequency within the range of [−π,π).
- The embedding of the watermark is performed as follows.
- y[n]=x[n]+w[n] (1)
- In the equation (1), the embedding of the watermark is performed in a manner that the watermark signal w[n] is added to the original signal x[n].
- In many cases, a receiver detects the watermark by calculating a correlation.
- Here, from a viewpoint of the watermark w[n], the original signal x[n] is considered as a colored noise, and the detection performance is greatly improved by the whitening filter.
-
- At this time, Rx is the covariance matrix of the original signal x[n].
- Meanwhile, decision of the test statistics by applying LRT in the hypotheses having the distribution of the above two equations is given by the following equation.
-
- is selected H1.
- Here, η is the detection threshold, and is decided by selection.
- In the above equation, if T(y) is larger than the detection threshold η, H1 is selected.
- Meanwhile, the Wiener attack and the power spectrum condition (PSC) will be explained.
- The Wiener attack is the removal of the estimated watermark, where the estimation is performed from the watermarked signal, in order to prevent the detection of the watermark.
- The attacked signal is given by
- ŷ[n]=y[n]−γŵ[n]=y[n]−γh[n]*y[n].
- In this equation, h[n] is the impulse response of the Wiener filter, and γ(≧0) is a scalar gain factor.
-
- In order for the watermark to be robust against the Wiener attack, the estimation of the watermark should necessarily be difficult.
-
- This equation means that the power spectral density of the watermark should be in proportion to the power spectral density of the original signal.
- As described above, however, in case of the watermark that satisfies the power spectrum condition (PSC), there is no detection gain of the whitening filter, and thus the detection performance is poor in comparison to the white noise watermark.
- On the contrary, if the power spectral density is different from the power spectral density of the original signal, the detection performance of the whitening filter improves, but the Wiener attack considerably removes the watermark from the received signal. Accordingly, the present invention provides a watermark designing method that satisfies the two conditions.
- FIG. 3 is a block diagram illustrating the Wiener attack and the whitening filter.
- y[n] denotes a received signal including the watermark w[n] and the original signal x[n], g[n] the impulse response of the whitening filter after the Wiener attack, and w[n] the watermark signal. A correlation detector performs a detection based on correlation between the two signals. After the Wiener attack, the signal passes through the whitening filter g[n], and then existence/nonexistence of the watermark is finally discriminated through the correlation detector.
- The optimum watermark designing in consideration of the whitening filter according to the present invention first maximizes the following equation.
- E[T 1(y)−T 0(y)] (2)
-
- In the equations (2) and (3), if the spectrums of the test statistics existing in the respective hypotheses are sufficiently separated, the detection performance is improved.
-
-
-
-
-
-
- That is, the watermark is designed so as to maximize the power spectral density Φx(Ω) of the original signal and the power spectral density Φw(Ω) of the watermark in the frequency range of [−π,π).
- In order to maximize the equation (6), the power spectral density of the watermark is first extracted using the Lagrange multiplier method.
-
- At this time, g(Φw(Ω)) is an integrand.
-
-
- The power spectral density Φw(Ω) of the watermark should be a positive real number. The power spectral density having a negative value is removed.
-
-
-
-
- That is, the optimum method of designing the watermark, which is robust against the Wiener attack and has an excellent detection performance even when the whitening filter is used, should satisfy the equation (8).
- The above equation (8) shows the power spectral density of the designed watermark w[n], and indicates that the power spectral density Φw(Ω) of the watermark is not in proportion to the power spectral density Φx(Ω) of the original signal. Also, the equation (8) represents a convex curve. If {square root}{square root over (Φx(Ω))} is too large, the power spectral density Φw(Ω) of the watermark is reduced on the contrary, and thus if the large power spectral density Φw(Ω) of the watermark is in proportion to the power spectral density Φx(Ω) of the original signal, the operating efficiency is not good.
-
- If the designed watermark does not satisfy the above condition, the optimum watermark cannot be obtained.
- FIGS. 4 and 5 are graphs of compared performances of the white noise watermark, the watermark satisfying the PSC, and the watermark provided by the designing method according to the present invention when the watermark document ratio (WDR)=−20 dB and WDR=−30 dB, respectively. In FIGS. 4 and 5, the optimum watermark corresponds to the watermark according to the present invention.
- FIGS. 4 and 5 illustrate the receiver operating characteristics, and show the relation between a probability of false alarm and a normal probability of detection. Under the same probability of false alarm, the performance becomes better as the probability of detection becomes higher. A better performance is shown under the ratio of the watermark to the original signal of −20 dB and −30 dB. As shown in the drawings, the WDR of the watermark designed by the optimization method between −20 dB and −30 dB represents an improved detection performance.
- FIG. 6 is a graph of a power spectral density Φx(Ω) of the original signal, and FIG. 7 is a graph of a power spectral density Φw(Ω) of the watermark according to the present invention.
- Generally, for a better detection through the whitening filter, the power spectral density of the watermark should be in high-frequency regions. However, as shown in FIG. 7, even if the power spectral density of the watermark is concentrated in low-frequency regions, a good detection performance is shown after the whitening filtering.
-
- In the above equation, T is the optimized test statistics.
- FIG. 8 illustrates the watermark according to the present invention showing the optimum result. There is not a great difference between the power spectral density of the watermark designed according to the present invention and the power spectral density of the original signal, but the performance of the whitening filtered detection improves.
- As described above, the watermark designed according to the present invention shows an improved detection performance in a detection environment using the whitening filter after the Wiener attack.
- If the power spectral density of the watermark is identical to the power spectral density of the original signal, the gain of the whitening filter becomes 1, and thus the detection performance is not improved. Also, if the power spectral density of the watermark is opposite to the power spectral density of the original signal, the watermark is easily removed by the Wiener attack even if the whitening filter has any gain, and this deteriorates the detection performance of the whitening filter. On the contrary, according to the present invention, the detection performance is improved using the watermark having the optimized power spectral density in consideration of both the Wiener attack and the whitening filter.
- While the present invention has been described illustrated herein with reference to the preferred embodiment thereof, it will be understood by those skilled in the art that various changes and modifications may be made to the invention without departing from the spirit and scope of the invention, which is defined in the appended claims.
Claims (4)
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR2002-37023 | 2002-06-28 | ||
KR10-2002-0037023A KR100449803B1 (en) | 2002-06-28 | 2002-06-28 | Watermark design for the Wiener attack and whitening filtered detection |
Publications (2)
Publication Number | Publication Date |
---|---|
US20040001607A1 true US20040001607A1 (en) | 2004-01-01 |
US7076081B2 US7076081B2 (en) | 2006-07-11 |
Family
ID=29774975
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/271,530 Expired - Fee Related US7076081B2 (en) | 2002-06-28 | 2002-10-17 | Method of designing watermark in consideration of wiener attack and whitening filtered detection |
Country Status (2)
Country | Link |
---|---|
US (1) | US7076081B2 (en) |
KR (1) | KR100449803B1 (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060015713A1 (en) * | 2004-06-03 | 2006-01-19 | Rhoads Geoffrey B | Economically secure digital mass media systems |
US20060124260A1 (en) * | 2004-12-10 | 2006-06-15 | Eastman Kodak Company | Bistable watermark |
WO2006065023A1 (en) * | 2004-12-13 | 2006-06-22 | Electronics And Telecommunications Research Institute | Method and systems for selecting test stimuli for use in evaluating performance of video watermarking methods |
US20060212710A1 (en) * | 2005-03-18 | 2006-09-21 | Thomson Licensing | Method and apparatus for encoding symbols carrying payload data for watermarking an audio or video signal, and method and apparatus for decoding symbols carrying payload data of a watermarked audio or video signal |
US20090235079A1 (en) * | 2005-06-02 | 2009-09-17 | Peter Georg Baum | Method and apparatus for watermarking an audio or video signal with watermark data using a spread spectrum |
CN115168915A (en) * | 2022-07-22 | 2022-10-11 | 浙江大学 | Self-adaptive watermarking method for integrity protection of real-time bus noisy data |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2003026153A1 (en) * | 2001-09-20 | 2003-03-27 | Exo-Brain, Inc. | Input-output device with universal phone port |
KR100653756B1 (en) * | 2006-04-21 | 2006-12-05 | 주식회사 용진유화 | Coating composition for green tires |
US7974437B2 (en) * | 2007-11-19 | 2011-07-05 | Seiko Epson Corporation | Identifying steganographic data in an image |
US8081823B2 (en) * | 2007-11-20 | 2011-12-20 | Seiko Epson Corporation | Segmenting a string using similarity values |
US8031905B2 (en) * | 2007-11-21 | 2011-10-04 | Seiko Epson Corporation | Extracting data from images |
US8243981B2 (en) * | 2007-11-26 | 2012-08-14 | Seiko Epson Corporation | Identifying embedded data in an image |
US8009862B2 (en) * | 2007-11-27 | 2011-08-30 | Seiko Epson Corporation | Embedding data in images |
US8798388B2 (en) * | 2009-12-03 | 2014-08-05 | Qualcomm Incorporated | Digital image combining to produce optical effects |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5848155A (en) * | 1996-09-04 | 1998-12-08 | Nec Research Institute, Inc. | Spread spectrum watermark for embedded signalling |
US5889868A (en) * | 1996-07-02 | 1999-03-30 | The Dice Company | Optimization methods for the insertion, protection, and detection of digital watermarks in digitized data |
US6332194B1 (en) * | 1998-06-05 | 2001-12-18 | Signafy, Inc. | Method for data preparation and watermark insertion |
US6535617B1 (en) * | 2000-02-14 | 2003-03-18 | Digimarc Corporation | Removal of fixed pattern noise and other fixed patterns from media signals |
US6625295B1 (en) * | 1996-09-13 | 2003-09-23 | Purdue Research Foundation | Authentication of signals using watermarks |
US20030182246A1 (en) * | 1999-12-10 | 2003-09-25 | Johnson William Nevil Heaton | Applications of fractal and/or chaotic techniques |
US20040006696A1 (en) * | 2000-06-08 | 2004-01-08 | Seung-Won Shin | Watermark embedding and extracting method for protecting digital audio contents copyright and preventing duplication and apparatus using thereof |
US6792129B1 (en) * | 1999-08-25 | 2004-09-14 | Sharp Laboratories Of America | Digital watermarking in a perceptually uniform domain |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR100611094B1 (en) * | 2000-06-15 | 2006-08-09 | 주식회사 케이티 | Apparatus and method for inserting/detecting watermark based stochastic model |
KR100896618B1 (en) * | 2002-06-27 | 2009-05-08 | 주식회사 케이티 | Apparatus and method for inserting and detecting digital image watermarking |
-
2002
- 2002-06-28 KR KR10-2002-0037023A patent/KR100449803B1/en not_active IP Right Cessation
- 2002-10-17 US US10/271,530 patent/US7076081B2/en not_active Expired - Fee Related
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5889868A (en) * | 1996-07-02 | 1999-03-30 | The Dice Company | Optimization methods for the insertion, protection, and detection of digital watermarks in digitized data |
US5848155A (en) * | 1996-09-04 | 1998-12-08 | Nec Research Institute, Inc. | Spread spectrum watermark for embedded signalling |
US6625295B1 (en) * | 1996-09-13 | 2003-09-23 | Purdue Research Foundation | Authentication of signals using watermarks |
US6332194B1 (en) * | 1998-06-05 | 2001-12-18 | Signafy, Inc. | Method for data preparation and watermark insertion |
US6792129B1 (en) * | 1999-08-25 | 2004-09-14 | Sharp Laboratories Of America | Digital watermarking in a perceptually uniform domain |
US20030182246A1 (en) * | 1999-12-10 | 2003-09-25 | Johnson William Nevil Heaton | Applications of fractal and/or chaotic techniques |
US6535617B1 (en) * | 2000-02-14 | 2003-03-18 | Digimarc Corporation | Removal of fixed pattern noise and other fixed patterns from media signals |
US20040006696A1 (en) * | 2000-06-08 | 2004-01-08 | Seung-Won Shin | Watermark embedding and extracting method for protecting digital audio contents copyright and preventing duplication and apparatus using thereof |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060015713A1 (en) * | 2004-06-03 | 2006-01-19 | Rhoads Geoffrey B | Economically secure digital mass media systems |
US9305148B2 (en) | 2004-06-03 | 2016-04-05 | Digimarc Corporation | Economically secure digital mass media systems |
US8667275B2 (en) * | 2004-06-03 | 2014-03-04 | Digimarc Corporation | Economically secure digital mass media systems |
US8334966B2 (en) * | 2004-12-10 | 2012-12-18 | Industrial Technology Research Institute | Method of forming a bistable watermark |
US20060124260A1 (en) * | 2004-12-10 | 2006-06-15 | Eastman Kodak Company | Bistable watermark |
US20090183824A1 (en) * | 2004-12-10 | 2009-07-23 | Industrial Technology Research Institute | Bistable watermark |
US8330932B2 (en) * | 2004-12-10 | 2012-12-11 | Industrial Technology Research Institute | Bistable watermark |
WO2006065023A1 (en) * | 2004-12-13 | 2006-06-22 | Electronics And Telecommunications Research Institute | Method and systems for selecting test stimuli for use in evaluating performance of video watermarking methods |
US20080085054A1 (en) * | 2004-12-13 | 2008-04-10 | Electronics And Telecommunications Research Institute | Method And Systems For Selecting Test Stimuli For Use In Evaluating Performance Of Video Watermarking Methods |
US8009859B2 (en) | 2004-12-13 | 2011-08-30 | Electronics And Telecommunications Research Institute | Method and systems for selecting test stimuli for use in evaluating performance of video watermarking methods |
US20060212710A1 (en) * | 2005-03-18 | 2006-09-21 | Thomson Licensing | Method and apparatus for encoding symbols carrying payload data for watermarking an audio or video signal, and method and apparatus for decoding symbols carrying payload data of a watermarked audio or video signal |
TWI403974B (en) * | 2005-03-18 | 2013-08-01 | Thomson Licensing | Method and apparatus for encoding symbols carrying payload data for watermarking an audio or video signal, and method and apparatus for decoding symbols carrying payload data of a watermarked audio or video signal |
US7634031B2 (en) * | 2005-03-18 | 2009-12-15 | Thomson Licensing | Method and apparatus for encoding symbols carrying payload data for watermarking an audio or video signal, and method and apparatus for decoding symbols carrying payload data of a watermarked audio or video signal |
US20090235079A1 (en) * | 2005-06-02 | 2009-09-17 | Peter Georg Baum | Method and apparatus for watermarking an audio or video signal with watermark data using a spread spectrum |
CN115168915A (en) * | 2022-07-22 | 2022-10-11 | 浙江大学 | Self-adaptive watermarking method for integrity protection of real-time bus noisy data |
Also Published As
Publication number | Publication date |
---|---|
KR100449803B1 (en) | 2004-09-22 |
KR20040001710A (en) | 2004-01-07 |
US7076081B2 (en) | 2006-07-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0941605B1 (en) | Method and arrangement for detecting a watermark | |
Cox et al. | Watermarking as communications with side information | |
Etemad et al. | A new multiplicative watermark detector in the contourlet domain using t location-scale distribution | |
Hernández et al. | Statistical analysis of watermarking schemes for copyright protection of images | |
Jabade et al. | Literature review of wavelet based digital image watermarking techniques | |
Su et al. | Novel approach to collusion-resistant video watermarking | |
Zou et al. | Steganalysis based on Markov model of thresholded prediction-error image | |
Chan et al. | A novel scheme for hybrid digital video watermarking: approach, evaluation and experimentation | |
EP1849124B1 (en) | System and method for steganalysis | |
US7076081B2 (en) | Method of designing watermark in consideration of wiener attack and whitening filtered detection | |
CN102063907B (en) | Steganalysis method for audio spread-spectrum steganography | |
WO2001054053A1 (en) | Transform domain allocation for multimedia watermarking | |
Maity et al. | Genetic algorithms for optimality of data hiding in digital images | |
Budhia et al. | Digital video steganalysis exploiting collusion sensitivity | |
JP3262159B2 (en) | Spread spectrum watermark processing method | |
Ng et al. | Wavelet domain watermarking using maximum-likelihood detection | |
Benseddik et al. | Efficient interpolation-based reversible watermarking for protecting fingerprint images | |
Chen et al. | Alpha-trimmed image estimation for JPEG steganography detection | |
Holliman et al. | Watermark extimation through local pixel correlation | |
Hsu et al. | New watermark-removal method based on Eigen-image energy | |
Hwang et al. | Watermark design for the Wiener attack and whitening filtered detection | |
Zebbiche et al. | Robust Multibit Decoding and Detection of Multiplicative Watermarks for Fingerprint Images. | |
KR100455062B1 (en) | Apparatus and method of embedding and detecting layered watermark | |
Jayalakshmi et al. | Optimum retrieval of watermark from wavelet significant coefficients | |
Barni et al. | Digital watermarking of visual data: State of the art and new trends |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTIT Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HWANG, YOUNGHA;LEE, KIRYUNG;MOON, KYUNG AE;AND OTHERS;REEL/FRAME:013397/0220;SIGNING DATES FROM 20020902 TO 20021002 |
|
FEPP | Fee payment procedure |
Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY |
|
FEPP | Fee payment procedure |
Free format text: PAYER NUMBER DE-ASSIGNED (ORIGINAL EVENT CODE: RMPN); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
FPAY | Fee payment |
Year of fee payment: 8 |
|
AS | Assignment |
Owner name: INTELLECTUAL DISCOVERY CO., LTD., KOREA, REPUBLIC Free format text: LICENSE;ASSIGNOR:ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE;REEL/FRAME:032456/0991 Effective date: 20140314 |
|
FEPP | Fee payment procedure |
Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.) |
|
LAPS | Lapse for failure to pay maintenance fees |
Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.) |
|
STCH | Information on status: patent discontinuation |
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 |
|
FP | Lapsed due to failure to pay maintenance fee |
Effective date: 20180711 |